diff --git a/src/diffusers/models/transformers/transformer_sd3.py b/src/diffusers/models/transformers/transformer_sd3.py
index ead657d0cfd2..9a56ca4e226d 100644
--- a/src/diffusers/models/transformers/transformer_sd3.py
+++ b/src/diffusers/models/transformers/transformer_sd3.py
@@ -289,6 +289,8 @@ def forward(
         height, width = hidden_states.shape[-2:]
 
         hidden_states = self.pos_embed(hidden_states)  # takes care of adding positional embeddings too.
+        # pos_embed output is non-contiguous due to flatten+transpose in PatchEmbed (BCHW -> BNC).
+        hidden_states = hidden_states.contiguous()
         temb = self.time_text_embed(timestep, pooled_projections)
         encoder_hidden_states = self.context_embedder(encoder_hidden_states)